# Real GenBank Benchmarks

The repository includes two small, fixed GenBank flat-file snapshots for
offline regression testing and manual profiling. They were downloaded from the
NCBI E-utilities `efetch` endpoint on 2026-08-11:

| Snapshot | Accession | LOCUS length | Feature coverage | SHA-256 |
| --- | --- | ---: | --- | --- |
| [`examples/u49845.gb`](examples/u49845.gb) | U49845 | 5028 | partial locations, `complement`, multiline translations, three CDS | `24065e39b5ab0543d34a5d893bda08ca6e9fdc2e26cb89430f0db9fb3af01ad0` |
| [`examples/af165912.gb`](examples/af165912.gb) | AF165912 | 5485 | `join(...)` across multiple lines, mRNA/CDS qualifiers, regulatory features | `e40d28295531872e498dee403fdc0eab0b8a7e94e3fa518ca6a911778f7a84b0` |

Primary sources:

- <https://www.ncbi.nlm.nih.gov/nuccore/U49845>
- <https://www.ncbi.nlm.nih.gov/nuccore/AF165912>
- <https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nuccore&id=U49845&rettype=gb&retmode=text>
- <https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nuccore&id=AF165912&rettype=gb&retmode=text>

These files are distributed as public database records for interoperability
testing. They are not MoonBit source code and are kept unchanged apart from
the downloaded file representation. The project does not claim ownership of
the underlying biological records; users should consult NCBI/GenBank terms
when redistributing or building larger datasets.

## Regression expectations

- U49845 must parse as one record with a 5028-base ORIGIN and one `source`
  feature. Its three CDS locations exercise partial and reverse-strand
  parsing.
- AF165912 must parse as one record with a 5485-base ORIGIN and a multi-line
  `join` location for both the mRNA and CDS features.
- Any change to these snapshots must update this manifest and record the new
  source URL, retrieval date, and SHA-256 value.